To the editor,

Therapeutic improvements over the past decades have made acute promyelocytic leukemia (APL) one of the most curable hematological malignancies [1, 2]. However, central nervous system (CNS) recurrences still occur. In the pre-arsenic era, the incidence of extramedullary relapse in APL approximately ranged from 3% to 5%, most commonly in the CNS [3,4,5,6]. Nevertheless, the prevalence and predictive factors of CNS recurrence in APL patients remain poorly defined in unselected populations during the arsenic era.

Here, we evaluated a large, consecutive cohort of newly diagnosed APL patients treated at our center over a 13-year period. We aimed to reveal the clinical epidemiology of CNS relapse and to identify the potential risk factors under real-world conditions.

Between January 2011 and December 2023, we enrolled consecutive patients with newly diagnosed APL admitted to our center. Six patients were excluded due to age <14 years (n = 3) or refusal of treatment at our center (n = 3). The diagnosis of APL was confirmed by the detection of the PML::RARA fusion gene. This study was approved by the institutional review board and conducted in accordance with the Declaration of Helsinki.

At the initial suspicion of APL, patients were promptly treated with all-trans retinoic acid (ATRA) at a dose of 25 mg/m2/d. Upon confirmation of PML::RARA, arsenic trioxide (ATO) at 0.15 mg/kg/d or Realgar-Indigo naturalis formula (RIF) at 60 mg/kg/d was administered concurrently with ATRA. Cytoreductive therapy was initiated immediately when white blood cell (WBC) counts exceeded 10 × 109/L to mitigate leukocytosis. Platelet (PLT) transfusions and fresh frozen plasma were routinely administered to maintain PLT counts above 30 × 109/L, and fibrinogen concentrations above 100–150 mg/dL.

For post-induction therapy, patients received three courses of anthracycline-based chemotherapy for consolidation, followed by sequential administration of ATRA and ATO or RIF as maintenance therapy from 2011 to 2015 [7]. After 2015, the ATRA plus arsenic chemotherapy-free strategy was adopted. Prophylaxis for CNS leukemia (cytarabine plus dexamethasone with or without methotrexate intrathecal) was routinely recommended at least three times for high-risk patients (WBC count >10 × 109/L) during consolidation.

Bone marrow(BM) relapse was defined as the presence of malignant promyelocytes in the BM exceeding 5% [8]. CNS relapse was diagnosed based on the detection of PML::RARA transcripts in the cerebrospinal fluid [9]. Overall survival (OS) was calculated from the first time of complete remission (CR) until death or the last follow-up. The follow-up was updated on June 1, 2024. Statistical analyses were performed using the Wilcoxon rank-sum test, Chi-square test or Fisher’s exact test as appropriate. Risk factors associated with CNS recurrence were evaluated using univariate and multivariate logistic regression analyses, with results expressed as odds ratios (ORs) and 95% confidence intervals (CIs). Kaplan-Meier analysis with the log-rank test was applied for survival analysis. All statistical analyses were performed by GraphPad Prism Software (version 9), and IBM SPSS Statistics Software (version 27) for Windows. A two-sided p-value < 0.05 was considered statistically significant.

A total of 910 newly diagnosed APL patients were eligible during the study period. Among these, 71 (7.8%) died during induction therapy, while the remaining 839 patients achieved CR. With a median follow-up of 74 months, 80 patients relapsed after CR, and the median time from diagnosis to relapse was 17 months (range 4–51 months). Baseline characteristics of the relapsed patients are summarized in Table 1.

Table 1 Baseline characteristics of the study population.
Full size table

CNS involvement at first relapse was documented in 13 patients, corresponding to an overall CNS relapse incidence of 1.4% (Table 2). Notably, 9 of these 13 patients developed isolated CNS recurrence, while 4 exhibited CNS relapse concurrent with BM relapse. The median time from diagnosis to CNS recurrence was 22 months (range, 7–39 months), 7 of 13 patients completed the scheduled arsenic-based treatments but developed CNS relapse (Supplementary Table 1). FLT3-ITD mutations were observed among 8 (61.5%) patients and 6 (46.1%) had a history of intracranial hemorrhage before or during induction (Table 2). The incidence of CNS relapse did not differ significantly between patients who received anthracycline-based chemotherapy consolidation and those who did not (2.1% vs. 1.1%; P = 0.384) (Supplementary Table 2). Furthermore, no statistically significant differences in baseline characteristics—such as age, gender, WBC count, PLT count, fibrinogen level, D-dimer level and FLT3-ITD mutation status—were observed across different time periods (Supplementary Table 2).

Table 2 Clinical features of relapsed patients.
Full size table

In addition, 67 patients experienced relapse without CNS involvement. Table 2 compares the baseline characteristics between patients with and without CNS relapse. Univariate analysis revealed that CNS relapse was associated with FLT3-ITD mutation (P = 0.008), intracranial hemorrhage before or during induction (P < 0.001) and fibrinogen level (P = 0.015). Multivariate analysis confirmed that intracranial hemorrhage (OR = 17.747, 95% CI:2.725–115.572; P = 0.003) and FLT3-ITD mutation (OR = 5.069, 95% CI:1.105–23.247; P = 0.037) were independent prognostic factors for CNS relapse (Supplementary Table 3). Other factors—including gender, age, WBC counts, hemoglobin level, PLT counts, Sanz risk category and D-dimer levels—did not show statistically significant differences.

OS was assessed for the 839 patients who achieved the first CR. The OS rate for patients with CNS relapse was significantly lower than that for relapsed patients without CNS involvement (47.2% vs. 83.5%; P = 0.0022) (Supplementary Fig. 1). As expected, both relapsed subgroups had a significantly worse prognosis compared to patients without recurrence (P < 0.0001).

To our knowledge, we analyzed the largest retrospective case series receiving modern arsenic-containing therapies. Our data indicate that early death and disease recurrence remain the primary modes of treatment failure in APL. Furthermore, the overall incidence of CNS relapse was 1.4%. Notably, our study demonstrated that consolidation treatment with or without anthracycline-based chemotherapy did not significantly affect CNS recurrence. Therefore, our findings provide evidence that a chemotherapy-free protocol does not increase the likelihood of CNS relapse.

The incidence of CNS relapse (1.4%) here corresponded with the prospective APL2012 trial (1.6%) in which all patients received ATO-based protocol [1]. Of note, prophylaxis for CNS leukemia was routinely performed in high-risk patients in both studies. Interestingly, in the APL0406 study, for low-to-intermediate risk patients, few CNS relapses occurred even without prophylactic intrathecal therapy [10]. The CNS relapse incidence in the arsenic era seemed comparable to that observed in previous pre-arsenic studies, which ranged from 0.6% to 4% [4,5,6, 11, 12]. Most of the pre-arsenic studies did not include prophylaxis for CNS leukemia. Thus, the clinical benefit of prophylactic intrathecal therapy remains to be confirmed.

In this study, the incidence of FLT3-ITD mutation was 61.5% among APL patients with CNS relapse compared to 20.9% in relapsed APL patients without CNS involvement. Similar findings were reported by Tashiro et al, who observed CNS involvement rates of 50% in the FLT3-ITD-positive group versus 5% in the FLT3-ITD-negative group [13]. It has been proposed that FLT3-ITD is associated with leukocytosis at onset and may promote leukemic infiltration of the CNS [14]. However, Tashiro et al also noted that two patients in the FLT3-ITD-positive group developed CNS involvement despite having low WBC levels [13]. In our study, 3 of the 8 CNS-relapsed patients with FLT3-ITD mutation were not classified as high risk according to the Sanz criteria. These findings, together with literature data suggest that FLT3-ITD mutation itself may serve as a predictor for CNS relapse in APL; however, further research is warranted to elucidate its role.

Another interesting finding is that the risk of CNS relapse increased progressively with the occurrence of intracranial hemorrhage before or during induction. Patients with CNS relapse exhibited a significantly higher rate of intracranial hemorrhage before or during induction compared to those without CNS involvement (46.1% vs. 3.0%). Multivariate analysis further confirmed the association between intracranial hemorrhage occurring before or during induction and subsequent CNS recurrence, underscoring its potential clinical relevance. Montesinos et al. [6] previously suggested that intracranial hemorrhage during induction may be a primary factor for CNS relapse. Consequently, proactive supportive therapies during induction—such as PLT and plasma transfusions—may not only reduce early death rates but also decrease the subsequent risk of CNS relapse.

Factors previously considered consistent predictors of CNS involvement, such as hyperleukocytosis [15] and younger age [12], showed limited predictive value in our study. This discrepancy may be attributable to differences in patient populations and clinical practices, as well as the relatively small number of CNS relapsed patients, which may introduce bias and affect statistical outcomes. Nevertheless, we recommend prompt control of hyperleukocytosis with cytotoxic agents to prevent differentiation syndrome. Further, larger prospective studies are necessary to improve the identification of patients at highest risk for CNS relapse.

This study has several limitations, primarily due to its retrospective design. Potential prognostic factors such as lactate dehydrogenase levels, differentiation syndrome and details regarding CNS prophylaxis were not included in the analysis, which may have influenced the multivariate analysis results. Moreover, given the extended duration of the study, not all patients received a uniform treatment regimen.

In conclusion, our analysis of a large cohort of APL patients treated with up-front arsenic-based therapy revealed a low incidence of CNS relapse. Furthermore, intracranial hemorrhage before or during induction and FLT3-ITD mutation status emerged as prominent risk factors for CNS relapse. These findings may help clinicians optimize APL management strategies to reduce the rate of CNS relapse and ultimately improve the overall cure rates in APL patients.